Chapter 25Language contact and loanwords

25.1 The Japanese stratum

Japonic contact with Ainu is the oldest documented layer and falls into two very unequal parts. A small, deep stratum entered before the island period: items such as sippo ‘salt’ ‘salt’ (Old Japanese shiho) and pasuy ‘chopsticks’ ‘chopsticks’ preserve features lost in later Japanese — the stop pronunciation of *p, the diphthong *uy — and so must have been received around the middle of the first millennium CE, when the ancestor of Ainu still lay in the immediate northern neighbourhood of the expanding Yamato state Janhunen (2022). These are pan-Ainu and not specifically Sakhalin; sippo is recorded in the earliest Hokkaidō and Kuril word-lists alike.

The bulk of the Japanese vocabulary is far younger and belongs to the Karafuto era of the Russo-Japanese frontier and the Japanese administration after 1905, when material culture flowed north in quantity. The diagnostic Sakhalin trait here is prosodic: the pitch accent of the Japanese source is rendered as vowel length, the Sakhalin counterpart of the Hokkaidō pitch contrast (Chapter 4 (Phonology)), so that an accented syllable surfaces long. The clearest set is the ‘metal’ family, all built on kaani ‘metal’ (< Japanese kane, with the long vowel for the accent), beside Hokkaidō kani with a short vowel Hattori (1964: 261).

Two of these words carry the chapter's recurring lesson about direction and route. tampaku ‘tobacco’ is formally a Japanese loan (tabako), but the same item travels the whole northern network — Nivkh tamχ, Uilta saŋna, Manchu dambaku (Shiraishi & Tangiku 2022) — so a Sakhalin speaker could have met it from more than one side. kanpi is the workhorse of the only substantial native-written Ainu prose we have, the Sentoku letters of 1906, where it means both the physical ‘paper’ and the ‘letter’ written on it; its first attested clause already turns on the wartime shortage of Russian paper:

Spoken code-switching into Japanese is also recorded, increasingly so in the latest generation of speakers. In Asai Take's post-war west-coast recordings, Japanese asides and discourse words are dropped into an otherwise Ainu grammatical frame — the narrator breaks into Japanese for an evaluative comment and returns to Ainu for the event line (Murasaki 2001). We treat such material as switching rather than borrowing, since the inserted Japanese keeps its own phonology and is not integrated into Ainu morphology (§25.6).

25.2 The Manchu–Amur trade stratum

For two centuries before the Russian and Japanese states arrived in force, Sakhalin hung off the northern end of the Manchu tributary-trade system that ran up the Amur: marten and sable pelts moved south, and silk, metalware, tea and spirits moved north, passing through Uilta and Nivkh hands before they reached the Ainu. The vocabulary of this trade is the clearest case in the language of borrowing by relay rather than face to face. Shiraishi and Tangiku's count is blunt about it: of the loanwords shared across the island's three languages, Nivkh and Uilta share by far the most (61 items), Nivkh and Ainu rather fewer (29), and Ainu and Uilta fewest of all (some 13) — a distribution that matches the historical record, in which the Ainu generally did not understand Uilta and communication ran one-way from the Uilta who knew Ainu (Shiraishi & Tangiku 2022). Direct Tungusic-to-Ainu borrowing is therefore thin; most Tungusic-looking words in Ainu arrived through the network, not from the Uilta next door.

The structurally most important members of this stratum are the borrowed counting-bases tanku ‘hundred’ ‘hundred’ and wantanku ‘thousand’, which match Manchu-Tungusic tanggū and underpin the productive decimal numeral system that Sakhalin Ainu developed for trade arithmetic; Piłsudski was told outright that tanku came ‘from the Olcha’ with the marten trade Piłsudski (1912: 141). Because they are already treated at length under the numerals, they are only cross-referenced here (Chapter 10 (Numerals and quantification)). The remaining trade words cluster in the expected domains — drink, ritual, status, and the exotic goods themselves.

The word nuca ‘Russian’ deserves a note because it is so often misattributed. It is not a borrowing from Russian; it comes from Orok/Uilta luča, with the initial l resolved to n by the Ainu avoidance of liquid onsets (§25.5), so that the name the Ainu used for the Russians reached them through their Tungusic neighbours rather than from the Russians themselves Chiri (1942: §17(2)). It is one of the highest-frequency loans in the corpus: every one of Sentoku's three letters turns on the comings and goings of the nuca (see (1)). The zither name tonkori is a second cautionary case — the instrument the Nivkh call tɨŋrɨŋ is organologically quite unlike the Ainu tonkori, so the name spread along the trade route even though the thing did not, and the two should not be equated (Shiraishi & Tangiku 2022). Finally, the west-coast mitten word wampakka ‘mittens’ is a genuine direct Uilta loan (Uilta wambakka), but a strictly local one: it is a west-coast isogloss against east-coast native matumere, treated under the dialects (Chapter 2 (Dialects, sources, and previous research)).

25.3 The Nivkh layer

The contact layer most likely to be under-represented in earlier grammars is the Nivkh one. Where the present description's predecessors cite barely a word or two, Shiraishi and Tangiku's appendix lists around twenty-nine items shared between Nivkh and Ainu, concentrated in the technologies of the central island — dog-sledding, sea-hunting, containers, and the fauna and tackle of the coast (Shiraishi & Tangiku 2022). The geography is telling: the Nivkh dialect richest in shared vocabulary is that of Poronaisk, in the central contact zone where Ainu, Nivkh and Uilta territories met, which marks the borrowing as local rather than island-wide.

The dagger marks the candour the data demand. For the dog-sled word, Chiri held nuso to be a borrowing from Nivkh nuci — there is no native Ainu word for the rig, and the nuci set is spread across the lower Amur — and Shiraishi and Tangiku accept this (Shiraishi & Tangiku 2022). But for tonto ‘tanned hide’ and upun ‘storehouse’ the likelier direction is the reverse, Ainu → Nivkh, so they are best not counted as Ainu loans at all. The general point is that for a great many of these the direction is simply debated; some may even come from a third, lost source — the language of the Okhotsk culture — that fed both Ainu and Nivkh Janhunen (2022). The reindeer word is the most famous instance of an old, deep exchange in this zone: Ainu tunakay ‘reindeer’ ‘reindeer’ goes back to Pre-Proto-Gilyak *tola-ŋa.y ‘pulling animal’, and was in turn passed on into Japanese as tonakai — Ainu here standing as the relay between Nivkh and Japanese, the mirror image of its receiving role elsewhere in the chapter Janhunen (2022), (Austerlitz 1976).

25.4 Russian and the Cyrillic register

Russian left Sakhalin Ainu almost no integrated lexicon. Such direct loans as are claimed are few and uncertain — kosika ‘cat’ (< koshka) is the stock example, but it is poorly attested and we flag it as unverified. The Russian contribution to the record is of a different order: it is a register and a script, not a word-list. Two things stand out.

The first is the Cyrillic writing tradition. Bronisław Piłsudski ran a short-lived east-coast school at which Ainu was written in the Russian alphabet, and the most substantial surviving body of Ainu prose by a native hand is the set of three letters that Sentoku Tarōji wrote him in Cyrillic in 1906, edited by Tangiku and Ogihara (Tangiku & Ogihara 2001). The orthographic side of this — how Cyrillic was made to register Ainu, and what it could not register (vowel length above all) — belongs to the orthography chapter (Chapter 3 (Orthography and transcription)), and the letters are glossed continuously in the texts chapter (Chapter 26 (Glossed texts)).

The second is the epistolary register the letters establish: an Ainu opening formula ituyaskara nispa ‘beloved master’ over a frame studded with Russian apparatus — calendar dates and place phrases (4go Fevral' ‘4th of February’, dating clauses on the Russian model), ruble-and-kopeck sums, and Russian personal and place names adapted to Ainu phonology: Korsakov, Aleksandr → Ariksandoru, lesnichiy ‘forester’ → Lesneciy. These are best read not as loans into the language but as the written furniture of a bilingual letter-writer; several further items that appear in the letters — atarimay, arenda ‘lease’, starosta ‘village elder’, menco, comen — we treat as unintegrated code-switches into Russian (or, for atarimay, Japanese) rather than established borrowings, and mark them accordingly. The point of substance is that Russian influence on Sakhalin Ainu is concentrated in this written, formulaic, name-and-number register, and barely touches the core lexicon at all.

25.5 Phonological adaptation of loans

Loanwords are reshaped to Ainu phonotactics by a small set of regular operations, most of them already visible in the tables above. They are synthesised here; the native phonological constraints they answer to are set out in Chapter 4 (Phonology), and only the contact-specific patterns are drawn together.

Liquid-onset avoidance. Ainu has no syllable-initial l, and a foreign initial liquid is repaired. The diagnostic case is nuca ‘Russian’ < Uilta luča, where l becomes n Chiri (1942: §17(2)). Accent → length. The pitch accent of a Japanese source is rendered as Sakhalin vowel length, the systematic correspondence behind kaani, tuuki against Hokkaidō short-vowel kani, tuki Hattori (1964: 117, 261). Voicing neutralisation. Ainu lacks a voicing contrast in stops, so voiced onsets of Japanese loans surface voiceless: beko → peko ‘cow’, buta → puta ‘pig’.

Geminate and cluster repair. A foreign geminate or illicit cluster is resolved into a permitted coda-plus-onset sequence, the coda surfacing as h by the regular coda neutralisation of the native phonology: thus Japanese teppō → tehpo ‘gun’, with hp for pp Hattori (1964: 118). Two of Chiri's showcase adaptations carry the repair further, through whole-segment loss and metathesis. The headwear word atampusa ‘a kind of hood’ comes from Japanese watabōshi 綿帽子 by way of an intermediate matampusi, then loss of the initial m Chiri (1942: §17(2); §19); and the lamp word, Hokkaidō ratcaku < Chinese 蠟燭 (via Japanese), appears in Sakhalin as cahraku ‘light, lamp’ by metathesis of the initial r Chiri (1942: §20).

The Nivkh t ↔ r correspondence. Across the Nivkh layer a Nivkh syllable-final t answers to an Ainu r: mut → muhru ‘pillow’, maχt → mahru ‘towline spreader’, motas → mohrasi ‘sled runner’ (Shiraishi & Tangiku 2022). The correspondence is regular enough to serve as a diagnostic of the layer, though it does not by itself fix the direction of any individual borrowing.

25.6 Morphological integration

The evidence for how loanwords behave in the grammar is thin, and we state only what the texts attest. The clearest fact is that integrated loan nouns are ordinary nouns: they take the native counting frames of Chapter 10 (Numerals and quantification) and slot into the same syntactic positions as inherited vocabulary. In the Sentoku letters the loan kanpi ‘letter’ is counted, possessed, and governed by postpositions exactly as a native noun would be, and it heads relative clauses (e=koro kanpi ‘your letter’ with the regular second-person possessive prefix). Trade-stratum nouns likewise take native quantifier and classifier frames: the borrowed counting-bases tanku and wantanku are not merely loaned words but loaned grammatical nouns, themselves countable with the native adnominal numerals (sine tanku ‘one hundred’), which is as deep an integration as a numeral system admits (Chapter 10 (Numerals and quantification)).

Whether loan verbs were taken in is harder to show. The integrated lexicon is overwhelmingly nominal — names of goods, animals, tools and materials — as is typical of trade borrowing, and the corpus offers no secure example of a borrowed verb stem inflecting with the Ainu personal prefixes. Where a foreign action concept is needed, the language reaches instead for a native verb or a light-verb construction with kara ‘make’ or ki ‘do’, leaving the loan in nominal shape. We flag this as an argument from silence rather than a demonstrated gap: the recorded corpus is small, and the latest speakers' free code-switching into Japanese (§25.1) shows that a foreign verb could always be reached for by simply switching language rather than by morphologically nativising the stem — which may be part of why a nativised loan-verb class never had to develop.